PDL::ImageND#
Image processing in any number of dimensions: convolution, rebinning, interpolation, circular masks, contours and graph layout forces.
use PDL;
use PDL::ImageND;
print convolveND(sequence(5), pdl(1, 1, 1)), "\n"; # [1 3 6 9 7]
print rebin(sequence(4), 2), " ", kernctr(sequence(4), pdl(1, 2, 1))->flat, "\n";
# [0.5 2.5] [2 1 0 1]
convolveND($a, $kernel, { boundary => 'periodic' }) convolves in as many dimensions as the kernel has, with a choice of boundary conditions and of the direct or the FFT method; convolve is the older form, wrapping around. rebin($a, @dims) changes the size by averaging or replicating ({ Norm => 1 } keeps the sum), ninterpol($point, $data, $n) interpolates linearly in the first $n dimensions, kernctr($image, $kernel) pads a kernel to the size of an image for the FFT, and circ_mean and circ_mean_p average around rings. contour_segments and contour_polylines trace a contour level through a 2-D field, path_join and path_segs turn edges into paths, combcoords packs three coordinates into one ndarray, and attract and repulse compute the spring and the repulsive forces of a graph layout.
This is PDL::ImageND of PDL 2.106: upstream’s Perl, unchanged, over PDL’s own kernels run by pperl’s engine.